The unmediated deployment of generative artificial intelligence in public search engines has surfaced a severe reputational hazard in Aotearoa. This week, an Auckland anaesthetic technician discovered that Google’s AI-generated search summaries had falsely branded her a criminal, laying bare the ongoing problem of algorithmic "hallucinations" and the lack of accessible recourse for ordinary citizens.
Tracey Melsom-Page, a healthcare professional with no criminal record barring the occasional speeding fine, was unexpectedly thrust into a digital nightmare. When her name was queried, Google's artificial intelligence falsely claimed she was linked to drink-driving convictions. The AI system had inexplicably conflated her identity with a completely unrelated case, fabricating a highly damaging narrative that appeared at the very top of the search interface.
The incident underscores a fundamental flaw in how large language models synthesise information. Rather than retrieving a specific, verifiable document, these systems generate text by predicting the next logical word based on vast, opaque datasets. When prompted to summarise information about an individual, the AI can seamlessly weave factual data with unrelated news reports or entirely synthetic claims, presenting the output with unwarranted authority and a factual tone.
Crucially, the burden of correcting these automated falsehoods currently falls squarely on the victim. Melsom-Page and her husband spent hours over the weekend attempting to navigate Google’s reporting channels to have the defamatory allegations removed. For professionals relying on a clean public record for employment, medical registration, and personal reputation, the latency in correcting such algorithmic errors can be profoundly damaging.
This local incident serves as a crucial warning for regulators and policymakers across New Zealand and the wider Asia-Pacific region. As major technology firms aggressively integrate generative AI overviews into default search functionality, the frequency of automated defamation is likely to increase. The situation raises urgent questions about platform liability under New Zealand’s existing defamation laws and the Harmful Digital Communications Act. If an AI agent synthesises a new, defamatory claim that was never published by a human author, the legal accountability for that publication remains dangerously ambiguous.
For enterprises and public institutions in Aotearoa, the Melsom-Page case is a stark reminder of the risks associated with deploying unchecked generative models for factual retrieval. Until technology companies can guarantee the provenance and accuracy of their AI summaries, or provide immediate, human-led mediation for false claims, the technology remains fundamentally unreliable for personal or professional queries. It is a clear signal that the pursuit of AI capabilities must be matched by equally robust safeguards for consumer protection.
Sources
- The New Zealand Herald: Google AI falsely links Auckland anaesthetic technician to drink-driving convictions
- Newstalk ZB: 'Ruining my life': AI falsely convicts anaesthetic technician of drink-driving
AI disclosure: Automated systems assisted with research and drafting. Gemini independently peer-reviewed this article before publication. Sources are linked for verification.
